30. The source of the lng rune and of the futhark*

Of all the 24 runes in the older juthark the 1J rune is by far the most disputed one. (Odenstedt 1990: 103)

Abstract

Adherents of traditional accounts of runic origins, namely of the Greek, Etruscan, and Latin theses, have been unable to identify the source of the 22nd rune, the g rune (named lng in the Old English rune poem), whose primary shapes in the older fupark are the diamond and the square, ¢, 0, and a problematic rounded variant, the circle, 0. In the present paper it is pointed out that all three basic shapes, including IJ and especially ¢, reflect attempts to represent a circle and that therefore all ancient Mediterranean alphabets offer a model, namely the circle letter, o. But whereas the sound value of 0 in the Greek alphabet as well as in its descendants, the Etruscan and Latin alphabets, is a vowel, [0], in the Phoenician alphabet and in the older fupark the circle letter represents consonants, namely a back voiced continuant consonant, l~] CAyin) in Phoenician, the velar nasal [lJ] in the fupark. Evidence is provided that substituting either a vowel or a nasal, especially a velar nasal, for r are typical ways of misperceiving and mispronouncing this Semitic sound. It is proposed that substituting a vowel became the Greek, substituting the velar nasal the Germanic way of adapting the Phoenician circle letter, CAyin. It follows in particular that only the assumption of a Phoenician source of the fupark can ex­ plain the sound value [lJ] of the lng rune in Germanic.

Zusammenfassung

Anhangern def traditionellen Auffassungen der Herkunft der Runen, namlich 00- griechischen, der etruskischen und der lateinischen These, ist es nicht gelungen, die Herkunft der 22. Rune des alteren Fuparks zu bestimmen, der g-Rune mit dem Namen lng im altenglischen Runengedicht und den primaren Formen ¢ und 0 sowie der problematischen Form o. 1m vorliegenden Artikel wird zunachst damuf hingewiesen, daB aile drei Primarformen, insbesondere IJ und vor allem ¢, aus dem Versuch resuitieren, einen Kreis wiederzugeben, und daB deshalb aile anti ken mit­ te!meerischen Alphabete ein Vorbild anbieten, namlich den Kreisbuchstaben, O. Aber wahrend der Lautwert von 0 sowohl im griechischen Alphabet als auch in seinen Abkommlingen, dem etruskischen und dem lateinischen Alphabet, ein 636 The source of the Ing rune

Vokal ist, namlich [0], reprasentiert der Kreisbuchstabe im phOnizischen Alphabet und im alteren Fupark einen Konsonanten, namlich einen stimmhaften hinteren konsonantischen Dauerlaut, [£] CAyin) im ersteren, [1)] im letzteren Schriftsystem. Es wird gezeigt, daB die Einsetzung eines Vokals oder eines Nasals, vor allem des velaren Nasals, fUr r charakteristische Fehlwahmehmungen und Falschaussprachen dieses semitischen Sprachlauts sind. Alsdann wird vorgeschlagen, daB die Ein­ setzung eines Vokals die griechische, die Einsetzung des velaren Nasals die ger­ manische Methode war, den ph6nizischen Kreisbuchstaben, CAyin, zu adaptieren. Daraus folgt insbesondere, daB nur die Annahme einer phOnizischen Herkunft des Fuparks den Lautwert [1)] der 22. Rune (<> etc.) im Germanischen erklaren kann.

30.1. The problem and a proposal

The Ing or +ingwaz rune, or U rune, of the older fupark has, among 1 others, the shape of a diamond, 0, and that of a square, 0. Since runes preferably consist of straight lines, shunning round forms, I consider 0 and 0 runic versions of a circle, 0, and call the lng rune the circle rune in certain contexts below. Since even horizontal lines are disfavored, the diamond is the proto-typical runic rendition of a circle. This is in har­ mony with modern textbook fuparks (e.g, Duwel 2008: 2), where the lng rune is represented by the diamond, and also with the following re­ sult of an intensive study of the various forms of the lng rune:

The most natural theory of the distribution of the variants is that <> (the variant in the fupark inscriptions) is the original U rune. (Odenstedt 1990: 118)

There are other opinions. In Krause 1966: 175 (also 1969: 160), the Opedal variant of the lug rune, a small circle with two straight lines pro­ jecting at the bottom, is considered the original form, "die vermutlich altertumlichste Gestalt, namlich die eines Kreises" ['probably the most archaic form, namely that of a circle'] (cf. Odenstedt 1990: 103),2 "Gr¢nvik ([1985:] pp. 13 184ff.) .. , agrees with Krause [1966, 1969], Antonsen r1982 J and other runologists and regards [the Opedal circle J as a malformed circle 0)" (Odenstedt 1990: 115), Odenstedt, however, does not reckon with carelessness: "Both lines below the 'circle' are so long and clear that it seems necessary to consider the possibility that Ithe Opedal circle] (not 0) is the intended form" (Odenstedt 1990: 109), Westergaard, who presents and discusses these and other shapes of the rune as well as their attestations (1981: 143- 156)3, does not consider any of the individual forms, 0, 0, or 0, as basic but all three of them, and in addition three further forms consisting of The source oj the Ing rune 637 the diamond with a vertical hasta at three different heights (Westergaard 1981: 173). The latter forms are considered by most runologists to be bind-runes combining the i rune I with the IJ rune ¢. The source of the lng rune has remained unexplained in the three major theories of runic origins, the Greek, Etruscan, and Latin theses. To Krause (1970: 40) the IJ rune is one of nine runes that are not "0 ber­ zeugend ableitbar" ['that have no convincing derivation']. Also more recent analyses of the runic system, or introductions to the runes, offer no acceptable explanations.4 Arntz (1944: 47) discusses the various shapes of the rune and con­ siders the square and diamond shapes as original. He writes:

Ein antikes Vorbild mUBte demnach etwa *0 gewesen sein; es ist aber nirgends als fJ belegt und nicht einmal wahrscheinlich zu machen. Keines der anti ken AI­ phabete hat ein besonderes Zeichen fur y; die Rune ist also mit einiger Wahr­ scheinlichkeit von den Germanen neugeschaffen. Ihr Lautwert ist nieht nur fJ, sondern auch yg und iyg. (Arntz 1944: 47)

['A model in one of the alphabets of antiquity would therefore have to have been something like *0; but it is nowhere attested, or even likely to be attested, as y. None of the ancient alphabets have a special sign for y; the rune was therefore with some probability newly created by the Early Germans. Its sound value is not only fJ but also yg and tyg. '] f4

It is not entirely clear why Arntz decided to look for a model * o. In his immediately preceding paragraph he says:

Die alteste Form scheint das ¢ von Vadstena, das vielleicht gerundct auf dem Stein von Arstacf und in wenig veranderter Form ... auf Opedal beJegt ist. Die gleiche Form kehrt, urn 90° gewendet, vielleicht auf Himmelstadlund wieder. Kylver zeigt ein liegendes Viereck 0. (Arntz 1944: 47)

['The oldest form appears to be the ¢ of Vadstena, which may be attested in rounded shape on the stone of Arstad and in only slightly modified shape ... on Opedal. Perhaps the same shape recurs, turned by 90", on Himmelstadlund. Kylver shows a lying rectangle 0.' j

It would appear more logical in view of this statement to look for a model * <> rather than * o. However, since both <> and 0 are runic ver­ sions of the circle both reconstructions are likely to be wrong. Rather, the model to look for in the alphabets of antiquity is *0. This conclu­ sion is quite independent of the question whether the Opedal quasi-cir- 638 The source oj the Ing rune

6 cle is a bona fide Ing rune or not. The rune shapes <.> and D by them­ selves tell us that they both derive from a circle of the source alphabet: The stylistic rules of rune design tell us that they both are circles, runic circles.7 Clearly then the proper model to look for in the possible Mediterranean source alphabets is +0. Looking for +0 in the Mediterranean alphabets - Phoenician, Greek, Etruscan, Latin is not a difficult task: They all have a letter of this shape, rs even Etruscan where the letter is not used in inscriptions but only in so-called model alphabets. Clearly therefore, the rune with the shapes <.>, D, and 0 is derived from the 0 of one of the Mediterranean alphabets of antiquity, and the only remaining question is from which. Evidently the shape of the rune is of no help, because the letter is a circle in all four possible source alphabets. Fortunately there is more to a letter than its shape. There is also its place in the alphabet and its sound value. As for the place of the 0 letter, it is unfortunately of no great help either. In the Phoenician alphabet it follows the sequence m n s: .8 In Greek, 0 follows m n x, x substituting for Phoenician s: . The Etruscan model alphabets show . When the Romans adopted the Etruscan alphabet, they had to select, for their own single voiceless sibilant phoneme 1sJ, one of three Etruscan letters for voiceless sibilant fricatives, s, s, and s, adaptations (via Gresk) of Phoenician s, ~, and s (with the Hebrew names Samrek" fiafie, and Sin). They opted for s. Since in particular s was not adopted, the Romans shortened the se­ quence to , which, with the vowel letter 0 for the 0, is also the order in the modem Western European alphabets, . In the fupark, 0 follows m I: ; I have proposed in Vennemann 2009b: 847f. that in the proto-fuJ:>ark the order was . If we knew that the Early Germans were good phoneticians we might guess that they simply interpreted the 0 letter as the third nasal, thus com­ pleting the subsequence phonetically as [m n 1)]. This may be suggestive, but it is also rather speculative. That leaves the sound value of 0 in the possible source languages as a criterion for identifying the source alphabet of the fupark. Here we do find a major difference. In the Greek alphabet, and therefore in the Etruscan and Latin alphabets, 0 was a vowel letter with the value [0] (which in Etruscan was not used because this language had no phoneme /0/, though possibly an allophonic variant fo] of the phoneme lui). In the Phoenician alphabet, 0 r6 was a consonant letter, as were all letters of that alphabet. In the fupark, the circle rune, <.>, D, 0, also was a consonant let­ ter. The source oj the lng rune 639

In Vennemann 2009b: 846f., I used the phonetic affinity between the circle letter in the Phoenician alphabet, where its sound value was [f phoneme and therefore could not adopt 0 with its original value. A third consideration was that the order of the runes in the proto­ fupark, where the circle rune probably followed directly after the nasals m and n, suggested to the Germanic alphabet adaptors that it be used as the third Germanic nasal, [IJ].

30.2. A new argument

Clearly then the consonantal value of the circle rune is more plausibly explained with Phoenician 0 (CAyin), a consonant letter, as a source than with Greek, Etruscan, and Latin 0, a vowel letter. But the match in the Phoenician case is even better than this. Jason Collett writes:

Henry Sweet and von der Gabelentz advised beginners in Arabic to substitute ng if they could not pronounce CAin. Interestingly, Hebrew CAyin is transcribed as a g, but with a small preceding superscript n, in Wigram's Hebrew Concordance aJthe Old Testament, in other words approximating CAyin as ng. (Jason Collett [Cape Town], e-Ietter to the author of 9 August 2009)

All three references lead to interesting results. Sweet wrote:

G. von der Gabelentz who united many of the qualifications of the theoretical and the practical linguist ... advises the beginner in Arabic who cannot

pronounce ... C en [i.e. cAin, T.V.] to substitute [lJJ - a sound which does not occur in Arabic, and therefore cannot be mistaken for anything but a substitute for (Gab. 75). Before I saw Gabelentz' book I had hit on the same device. (Sweet 1900: 35)

"Gab. 75" here refers to von der Gabelentz's book Die Sprachwissen­ schaft of 1891. I read in the reissued 2nd edition: f7

Wer Sprachen aus Buchem erlemt, thut gut, die fremden Laute, zumal die Vo­ cabeln und Pamdigmen, vemehmbar auszusprechen. Beabsichtigt er nicht mundlich in der fremden Sprache zu verkehren, so mag er die seinem Organe schwierigen Laute durch bequemere ersetzen, z.B. ... das semitische ' Ain [recte CAin, T.V.] wie ein h (= ng)9 aussprechen: arabisch Jahala, Janalat, jafoala, 640 The source of the Ing rune

fuhila [Le. facala,Jacalat, praia, fuci/a, forms of the verbal root pol 'to do, make. act', T.V.] .... So wirken Ohr und Auge zusammen, um die lastige Gedachtnisarbeit zu fordern. (von der Gabelentz 1901: 73)

[' When learning languages from books it is good practice to pronounce the for­ eign sounds, especially the words and paradigms, perceptibly. If there is no in­ tention to communicate orally in the foreign language, the sounds difficult for one's speech organs may be replaced by more comfortable ones, e.g. the Se­ mitic CAin by h (= ng): Arabicfanala,janalat,jajhala,fuhila [i.e. facala,Jacalat, )alaia,fucila, forms of the verbal rootP-1 'to do. make, act', T.V.]. n. In this way the ear and the eye cooperate to further the cumbersome work of memoriza­ tion. 'J

Collett's mention of Wigram's Hebrew Concordance of the Old Testa­ ment refers to the following passage (plus the transcription practice throughout the book):

''The system of pronunciation adopted requires little explanation: the simple English letters have been used so far as their sounds express those of the [He­ brew] original. The gutturals are thus represented, ... ; [CAyinJ by "g." (Wigram 1996: xviii)

This transliteration clearly has a basis in phonetic observation. That is evident from the following description on the website Ayin:

In some historical Sephardi pronunciations, cayin represented a velar nasal ([I)]) sound, as in English singing, while in non-' Mizrahi , modem Israeli Hebrew [it] represents a glottal stop in certain cases, but is mostly silent (that is, it be­ haves the same as aleph). However, often changes in adjoining vowels testify to the former presence of a pharyngeal or epiglottal articulation. 1O r8

I found indirect confirmation of the reality of the mispronunciation of CAyin, especially the Arabic CAin, as [lJ] in Wright's Grammar oj the Arabic language:

CAin, the Hebrew CAyin is a strong (but to [most] Europeans, as well as Turks

and Persians, unpronounceable) guttural. n. It is described as produced by a smart compression of the upper part of the windpipe and forcible emission of the breath. It is wrong to treat it, in any of the Semitic languages, as a mere vowel letter, or (worse still) as a nasal nor ng. (Wright 1896: 6) The source afthe Ing rune 641

Needless to say a teacher's no-no presupposes precisely what his admo­ nition is supposed to prohibit: Clearly Professor Wright (1830 - 1889) was in his days confronted with exactly these two mispronunciations of Arabic CAin and Hebrew CAyin by his students, as a vowel and as a na­ saL ll We may safely conclude that the Ancient Greeks and the Early Germans were simply the first offenders, the former substituting a vowel, the latter the velar nasal. It is, however, possible that the Early Germans were not the only adaptors of the alphabet to interpret the sound of CAyin as the velar na­ sal. In the Uigur and Yenisei variety of the Turkic "runic" script, and in the manuscripts found in Eastern Turkestan, 0 and ~ are used to repre­ sent lJ (Tekin 1968: 24 f., n. 2). The circle with the dot inside is the original form of the letter CAyin 'eye', the dot representing the pupil; but it may simply have been reinserted ornamentally, as probably in the Northern Iberian script where Phoenician CAyin takes the shapes 0, 0, and I:J (Jensen 1969: 282). The origin of the Turkic runic alphabet too is debated, but interestingly its decipherer proposed a Semitic alphabet as its source: r9

La source d'ou est tin~e l'origine de l'alphabet turc, sinon immediatement, du moins par intermediaire, c'est la forme de I'alphabet semitique qu'on appelle arCll11ienne. 12 C' est ce que prouvent quantite de ressemblances speciales dans la forme et la signification des lettres, outre que la direction de l'ecriture de droite a gauche concorde aussi particulierement bien avec cela. (Thomsen 1896: 46f.)

I The source from which the alphabet derives if not immediately, then at least via intermediary is the form of the Semitic alphabet called Aramaic. That is proved by the number of specific resemblances between the shapes and the meanings of the letters, apart from the fact that the direction of writing from right to left too agrees particularly well with this.']

Just like the Germanic runes, the Turkic runes were used for several centuries (Thomsen 1896: 54) until they fell into disuse.

30.3. Why a rune for sub-phonemic [lJ1?

The above identification of the source of the 9 rune, together with the reconstruction of the phonetic motivation for the adaptation of this source, /fj/, as [lJ]' may be suited to answer the question of why the Early Germans carried a letter in their writing system that represented a sound that by nearly all modern phonological analyses was not phonemic in 642 The source of the Ing rune their language. Ball (1988) has made a remarkable suggestion in answer to this question, one which he himself and his readers probably saw as indemonstrable at the time. Discussing the two occurrences of the g rune on the Ruthwell Cross, he wrote:

Like other runes it takes its value from its name (lng, [ilJg)). That it exists at alI is something of a puzzle, since it serves to distinguish the velar from the al­ veolar nasal sounds which apparently were not phonemicalIy distinct in Old English (or in Germanic) at all. ... It may be, perhaps, that the runic script was first adopted from a people whose language had a phonemic IIJ/. (Ball 1988: 114)

This surprising suggestion - surprising because none of the traditionally contemplated source languages of the fupark possess a phoneme IgI­ may be considered to have found an equally surprising indirect confirmation rIO through the above account: The source of the g rune was indeed phonemic in the language of the people from whom the runic script was first adopted, even though this source, fil, was not pho­ netically [IJ] but a sound that could be misunderstood and mispro­ nounced as l1)).

30.4. Conclusion

The g rune of the older fupark is by its shape (0, D, and the Opedal cir­ cle) and by its original place after the nasals identified with the 0 letter of the ancient Mediterranean alphabets. The letter 0 represents a vowel in the Greek alphabet and its derivatives, the Etruscan and Latin alpha­ bets, but a consonant in the Phoenician alphabet and - in its "runified" forms, D and especially 0 in the Germanic older fupark: a voiced pharyngeal fricative in the former, a velar nasal in the latter, hence back voiced continuant consonants in both writing systems. My earlier conclusion that the g rune cannot therefore be an adaptation of the Greek, Etruscan, or Latin 0 but only of the Phoenician 0 is here sup­ ported with a new argument. namely that non-native learners of Arabic or of Biblical Hebrew tend to substitute either a vowel or a nasal, espe­ cially the velar nasal, for the unfamiliar sound of Arabic CAin, Hebrew CAyin, /~/. It is here proposed that that is exactly what happened in the creation of the western European alphabets, the Greeks adapting Phoe­ nician 0 (with its sound value I~n as a vowel, the Early Germans as a nasal, their velar nasal. Evidently therefore only the Phoenician theory of runic origins allows an explanation for the Germanic assignment of The source oj the log rune 643 the consonant (1)1 to the 22nd runic letter (0, 0, Opedal 0), and for naming it +ingwaz.

30.5. Appendix: Odenstedt's Q => 0 proposal

The best proposal within the traditional theories for the source of the g rune 0 that I have seen is in Odenstedt's book. Whereas other runolo­ gists see <> as a Germanic innovation, a "coinage" (Odenstedt 1990: 160), Odenstedt himself considers it "possible to derive this rune from [the Latin capital letter] Q" (Odenstedt 1990: 161). As far as I can see, this proposal has not been appreciated even by adherents of the Latin thesis of runic origins, and I can understand why: The shapes of the circle rune do not express the rl1 asymmetry inherent in the letter QI3, and the sound value of Latin Q, [kw], is not a very good phonetic match for that of the circle rune, [lJ]' E.g., Spurkland (2005: 6), 15 years after Odenstedt's proposal, simply lists the g rune as one of twelve whose forms are "unknown in Roman script" and may therefore be owed to "other influences making their mark" or to the alphabet inventors' "imaginative approach to composing their vernacular alphabetic script". DOwel, an advocate of the Latin thesis (cf. DOwel 2008: 181), lists Odenstedt's book in his bibliography but does not mention his proposal for the origin of the Ing rune. 14 But in a letter of 31 October 2009 DOwel asks whether this special rune might not have been created (" geschaffen"), starting from the name *Ingwaz, specifically to indi­ cate ("kennzeichnen") the god named Yngvi in Old Norse, lng in Old English. He suggests that such a connection is thinkable ("denkbar") because of the acrophonic principle, nasal + consonant being syllabising ("silbebildend") anyway, hence ng = ing. In my opinion Odenstedt's proposal is not so bad that it should be ignored by the defenders of a Latin origin of the fupark. It may even be improved upon in such a way that the slight difference between Q and 0 in the Latin alphabet, the little "tail" of the Q, finds a certain cor­ respondence in the fupark. Considering the diamond form of the g rune as basic, the 0 rune only differs from it by having the two excrescences at the bottom, making it look like a stylized Greek omega. Thus, the 0 rune is "tailed", too, even though it has two tails rather than one. Fur­ thermore, Q and 0 are only separated by P in the Latin alphabet. In the fupark, the g rune and the 0 rune are separated by the d rune on the Kylver stone (Dtiwel and Heizmann 2006: 9); but they are adjacent on the Grumpan bracteate (DOwel and Heizmann 2006: 6) and on the Vadstena bracteate (DOwel and Heizmann 2006: 13). Assuming that the 644 The source oj the Ing rune latter fupark order, that with the subsequence , is the more original one, it seems possible that the shapes of the original g rune, the "tailed" diamond, and the original 0 rune, the plain f12 diamond, were confused, with the result that now the g rune is the plain diamond and the 0 rune, the "tailed" diamond. The "tailed" Opedal circle for g may hark back to the time before the confusion, and so may the English g rune, a dia­ mond which is doubly "tailed", at the top and at the bottom. - Even with this additional reconstructive step Odenstedt's reconstruction is not really convincing: On the formal side there remains the symmetry problem, and on the phonetic side nothing has changed. Since I have proposed my own solution to the problem of the source of the lng rune, I have no reason to defend Odenstedt. But his letter­ similarity approach to the problem of the lng rune seems to me to be closer to the way alphabets are normally designed on given models than Spurkland's unspecific "other influence" and "imagination" ap­ proach and Dtiwel's "creation" approach. In closing I would like to point out that although probably wrong, Odenstedt's proposal Q :=> <> supports my reconstruction 0 :=> <> on the formal level by interpreting the diamond rune as a runic circle, derived from a circular model, even if a "tailed" one.

Notes

First published 2010 in Sprachwissenschajt 35, 1-14. I would like to thank John Ole Askedal (Oslo), Klaus Dtiwel (Gottingen), Lutz Edzard (Oslo), Stephen Laker (Leiden), Angelika Lutz (Erlangen), Robert Mailhammer (Canberra), Patrizia Noel (Bamberg), and Karl Reichl (Bonn) for valuable suggestions. They are not responsible for the views expressed here. I. Much of what has been written about this rune appears to be rather uncertain (cL Birkmann 1995: 13-15). But nearly all specialists would subscribe to

the interpretation that the older fupark included a rune with the shapes [J (Kylver stone) and <> (Vadstena bracteate) and a sound value of [1]], or con­ taining [1]] (cf. especially Antonsen 2002: lO2f. for the latter view, and Westergaard 1981: 160-168 for probable phonetic correlates of all individual occurrences of the g rune). The interpretation below rests on these assump­ tions. 2. For this reason, and to facilitate printing, I represent the Opedal variant as a circle, o. The source of the Ing rune 645

3. Cf. the slightly different tabulation of the runic objects and their dates in Odenstedt 1990 (on page 107 for tJ and on page 108 for itJ). 4. Two proposals are discussed in the Appendix below. 5. More recent readings take this rune to be the k rune rather than the 9 rune; cf. Birkmann 1995: 13-15, Antonsen 2002: 4f. 6. Odenstedt considers the Opedal circle an "aUograph" of one of the bind runes for itJ. "Whatever the explanation of [the Opedal circle] it is clear that the sound value is itJ, and it is more than doubtful that it should be equated with Kylver's <> [sic, for 0, T.V.]" (Odenstedt 1990: 110). 7. The attestations themselves do not suggest any particular direction of de­ velopment for the shapes of the 9 rune, as Westergaard (1981: 168-172) shows. He says in particular (1981: 174) that assuming the circle as a start­ ing point, though very attractive, has no justification in the data. Even the single circle form on record is not perfect, as already mentioned above: The g rune of Opedal resembles a circle drawn by starting at the bottom line arxl returning to the bottom line, with the beginning and the end of the dra'wing crossing, thus creating two excrescences or tails below the circle (Wester­ gaard 1981: 174). In my view all this does not affect the argument; I would consider 0, and especially <>, runic adaptations of a circle even if no rounded forms at all existed for the 9 rune. 8. In order to facilitate printing and reading, Phoenician, Greek, Etruscan, arxl Roman letters (except for the circle letter, 0) are here transliterated with Latin minuscules. 9. The diacritic letter h is used by Indologists to transliterate the Sanskrit pho­ neme /rJ/. 10. Cf. the website "Ayin", http://en.wikipedia.org/wiki/Ayin (10 November 2009). An interesting discussion of the pronunciation of CAyin as [IJ] in Yiddish, in Judeo-Italian, and in the pronunciation of Hebrew by Dutch-Por­ tuguese and Italian Jews may be found on the website "Jewish-Languages Mailing List", December 2001, http://wwwjewish-languages.Qrglmll 2001 12.html (10 November 2009), especially in the contribution there by Prof. Gideon Goldenberg (Jerusalem). For a survey of the nasal pronuncia­ tion ([IJD of CAyin in (a part of) the Ashkenazic as well as the Italian arxl Portuguese Jewish communities cf. the lemma "Hebrew Grammar" in the Encyclopedia Judaica, col. 85-86. 11. Interestingly modern phonetic analyses confirm that pronunciations of CAin may be nasalized, cf. Kastner 1981: 49 and Hetzron 1969, especially the contribution there titled "Professor [Pierre C.] Delattre's statement", pp. 72- 73. 12. The oldest form of the Aramaic alphabet is essentially the same as the Old Phoenician alphabet; cf. Jensen 1969: 291. 646 The source of the Ing rune

13. The symmetrical shape of Greek and Etruscan Q only occurs in Archaic Latin inscriptions. In Classical Latin times - indeed as early as the funerary inscriptions of the Scipiones (3rd century, cf. Jensen 1969: 517, Wallace­ Hadrill 2008: figure 14) the letter Q has the asymmetrically tailed form still in use today. 14. This is in line with the design of the book, in which no attempt is made to trace the individual runes to specific sources.